distilgpt2-finetuned-scificorpus
This model is a fine-tuned version of distilgpt2 on the scifi-corpus dataset from kaggle: https://www.kaggle.com/datasets/jannesklaas/scifi-stories-text-corpus. It achieves the following results on the evaluation set:
- Loss: 5.0848
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 10
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 6.019 | 1.0 | 1520 | 5.8959 |
| 5.7047 | 2.0 | 3040 | 5.6273 |
| 5.5153 | 3.0 | 4560 | 5.4587 |
| 5.3785 | 4.0 | 6080 | 5.3378 |
| 5.2763 | 5.0 | 7600 | 5.2498 |
| 5.1943 | 6.0 | 9120 | 5.1881 |
| 5.139 | 7.0 | 10640 | 5.1430 |
| 5.0951 | 8.0 | 12160 | 5.1096 |
| 5.0665 | 9.0 | 13680 | 5.0923 |
| 5.0423 | 10.0 | 15200 | 5.0848 |
Framework versions
- Transformers 4.40.1
- Pytorch 1.13.1+cu117
- Datasets 2.19.0
- Tokenizers 0.19.1